Related Experiment Video
Updated: Jan 18, 2026
Tracking the Mammary Architectural Features and Detecting Breast Cancer with Magnetic Resonance Diffusion Tensor Imaging
Published on: December 15, 2014
Identification of selection criteria for DIBH in patients with left breast cancer using 3D-CRT and VMAT: An analysis
H Ayouni1, S Zarraa2, B Daoud3
1Higher Institute of Medical Technologies of Tunis, University Tunis El Manar, 1006, Tunis, Tunisia; Department of Medical Physics, Salah Azaiz Institute, 1006, Tunis, Tunisia.
Introduction:
Deep-inspiration breath-hold (DIBH) is a well-established technique for sparing organs-at-risk in breast cancer (BC) radiotherapy. Nonetheless, consensus on who benefits most remains elusive. This study aimed to establish dependable thresholds for anatomical parameters to identify these patients during three-dimensional conformal radiotherapy (3D-CRT) and volumetric modulated arc therapy (VMAT), and to confirm selection using multivariate logistic regression (MLR).
Methods:
Forty-eight left BC patients were included in group A, subdivided into 20 undergoing breast-conserving surgery (group B) and 28 undergoing radical mastectomy (group C). Free-breathing 3D-CRT plans were replanned using DIBH, each plan generated using VMAT. Ratios of heart volume in field (%HVIF) and left lung volume (%LLV) between free-breathing/DIBH defined eligibility thresholds. Spearman correlation assessed relationships between %LLV/%HVIF and dosimetric parameters or normal tissue complication probabilities (NTCPs). MLR confirmed selection, and models' performance was evaluated using area under curve (AUC).
Results:
Thresholds were 1.7 for %LLV and one for %HVIF. In 3D-CRT, %LLV significantly correlated with cardiac/left anterior descending artery (LAD) doses and LAD NTCP (groups A/C), while only mean heart dose (ΔMHD) and cardiac ΔV25 demonstrated significance (group B). In VMAT, %LLV significantly correlated with ΔMHD, heart maximum dose (ΔDmax), LAD mean dose (ΔDmean), and LAD ΔNTCP, independent of group. For %HVIF, significant correlations were found with heart %V25/%V40 (groups A/B) and LAD %Dmean (group A) in 3D-CRT but not in VMAT or group C. Thresholds of %LLV predicted DIBH with AUCs of 0.85/0.90/0.94 (3D-CRT) and 0.77/0.91/0.89 (VMAT). A similar trend was observed for %HVIF: 0.88/0.93/0.95 (3D-CRT) vs. 0.70/0.90/0.81 (VMAT).
Conclusion:
%LLV is a useful tool for selecting DIBH across techniques, while %HVIF is more applicable to 3D-CRT.
Implications For Practice:
This streamlined approach supports more personalized clinical decision-making.
